- Email: [email protected]
Inhibition of invasion and experimental metastasis of murine melanoma cells by human soluble thrombomodulin
Cancer Letters 161 (2000) 231±240
www.elsevier.com/locate/canlet
Inhibition of invasion and experimental metastasis of murine melanoma cells by human soluble thrombomodulin Yoshitaka Hosaka, Toshiyuki Higuchi, Michiko Tsumagari, Hidemi Ishii* Department of Public Health, Showa Pharmaceutical University, Higashi Tamagawa Gakuen, Machida, Tokyo 194-8543, Japan Received 24 July 2000; received in revised form 21 September 2000; accepted 22 September 2000
Abstract Thrombomodulin (TM) is an anticoagulant molecule expressed on the endothelial cell surface and soluble TM antigen, which is present in human plasma and urine, represents the products of limited proteolytic cleavage of cell-surface TM. Recently, it was demonstrated that TM is also expressed on the surface of several tumor cells and the expression level of TM negatively correlated with malignancy in cancer. We investigated the effect of soluble TM isolated from human urine (uTM) on the invasion and metastasis of murine melanoma cells (B16F10 cells) through a reconstituted basement membrane (Matrigel) and in a murine model of experimental lung metastasis. Matrigel reconstituted with uTM inhibited the invasion of B16F10 cells in a dose-dependent manner in a range from 10 to 1000 ng/ml uTM as compared with the control Matrigel without uTM. The inhibitory action of uTM was not altered in the presence of an excess amount of hirudin, an inhibitor of thrombin proteolytic activity, but abolished in the presence of anti-human TM IgG. Matrigel reconstituted with thrombin (1 NIH unit/ml) enhanced the invasion level of cells by 1.5-fold relative to the control Matrigel without thrombin. The thrombin-enhanced invasion of B16F10 cells was repressed by addition of hirudin (10 units/ml) or uTM (100 ng/ml) into the Matrigel. Matrigel reconstituted with hirudin (10 units/ml) and uTM (100 ng/ml) additionally accelerated the inhibitory activity of hirudin or uTM on the thrombin-enhanced invasion of B16F10 cells. Moreover, metastatic colonies formed in the lungs of mice injected intravenously with B16F10 cells were signi®cantly reduced by injection of uTM once a day up to 2 days after co-injection of uTM with the cells. These results suggested that Matrigel reconstituted with uTM inhibited the invasion of B16F10 cells in vitro through a thrombin-independent mechanism and the injection of uTM suppressed experimental lung metastasis of the cells in mice. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Thrombomodulin; Thrombin; Invasion; Metastasis
1. Introduction Tumor metastasis is a complex series of multi-step processes [1] of which invasion of tumor cells through the basement membrane is a critical step in the formation of metastases [2,3]. Tumor invasion consists of at * Corresponding author. Tel.: 181-42-721-1561; fax: 181-42721-1561. E-mail address: [email protected] (H. Ishii).
least three steps, which include attachment of tumor cells to matrix components, local degradation of matrix by tumor associated proteases such as urokinase-type plasminogen activator (uPA) and matrix metalloproteinases (MMPs), and locomotion of tumor cells into the region of the matrix degraded by pericellular proteolysis. Furthermore, in the case of blood-borne metastasis, thrombotic events play an important role in the pathogenesis of metastasis, since the formation of tumor emboli which arises from acti-
0304-3835/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(00)00617-0
232
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
vation of the clotting cascade and platelet aggregation facilitates the attachment and invasion of tumor cells through endothelial cells and the subendothelial matrix [4±6]. Thrombomodulin (TM) is a thrombin receptor found on the surface of vascular endothelial cells. Forming a high-af®nity complex with thrombin, TM directly inhibits the procoagulant activity of thrombin such as ®brin formation and accelerates the thrombindependent activation of the anticoagulant protein C zymogen [7]. Activated protein C inactivates coagulation factor Va and VIIIa in the presence of cofactor protein S which results in reduced thrombin formation [7]. Therefore, it has been postulated that TM is an important natural regulator in maintaining the ¯uidity of circulating blood [7]. Human TM consists of ®ve domains: the lectin-like domain (amino-terminal domain), epidermal growth factor (EGF)-like domain, O-glycosylation site-rich domain and transmembrane and cytoplasmic domains [8,9]. The EGF-like domain is composed of six tandem EGF-like structures having homology with the domain-III of the human EGF precursor [8,9]. The fourth, ®fth and sixth EGF-structures and the chondroitin-4-sulfate glycosaminoglycan bound O-glycosylation site-rich domain participate in the cofactor activity of TM for thrombin-catalyzed activation of protein C [10,11]. TM antigen is found not only on the endothelial cell surface, but also in human plasma and urine as soluble forms, which probably represent the products of limited proteolytic cleavage of cell-surface TM [12± 15]. We isolated the major active form of naturally existing soluble TM from human urine (uTM) which migrated as two bands of 72 and 79 kDa contained the lectin-like domain, the EGF-like domain and part of the O-glycosylation site-rich domain, but lacked the transmembrane and cytoplasmic domains of native TM [16±18]. The inhibitory activity of uTM for procoagulant activity of thrombin and the cofactor activity for thrombin-catalyzed protein C activation are retained. Therefore, uTM exhibits strong anticoagulant activities in vitro and in vivo [16,17]. TM also inhibits ®brinolysis through the acceleration of thrombin-catalyzed inactivation of single chain uPA (latent form of uPA) [19] in vitro and the activation of the thrombin-activatable ®brinolysis inhibitor (TAFI) in a puri®ed system [20] and in plasma [21]. Therefore, TM participates in the regulation of ®brinolysis and
the prevention of tissue damage under conditions where coagulation and/or ®brinolysis is accelerating. Recently, it has been reported that TM is present in several tumors, such as hepatocellular carcinoma [22], ovarian cancer [23] and squamous cell carcinoma [24], and that there is a negative correlation between the expression level of TM and malignancy of the cancer. We also suggested that TM may serve as a prognostic factor for metastatic breast cancer, since patients with low TM expression in breast cancer tissues had a signi®cantly lower rate of disease-free survival than patients with high TM expression [25]. Therefore, it was suggested that TM may play a role as a regulator during tumor invasion and metastasis. However, there is no experimental evidence that TM has inhibitory action against tumor invasion and metastasis. In the current study, we investigated the effects of uTM, a major active form isolated from naturally existing soluble TM antigen in human urine, on invasion and metastasis of murine melanoma cells (B16F10 cells) using reconstituted basement membrane (Matrigel) and an experimental lung metastasis model in mice.
2. Materials and methods 2.1. Materials Reagents were purchased from Wako Pure Chemical Industries (Osaka, Japan), except where indicated. Thrombin, hirudin, ecarin (Echis carinatus venom) and bovine serum albumin (BSA) were obtained from Sigma (St. Louis, MO). Chromogenic substrate S-2238 was obtained from Daiichi Chemical Co. (Tokyo, Japan). Prothrombin was from Enzyme Research Laboratories (South Bend, IN). Polyclonal rabbit anti-human TM IgG was produced by immunization of rabbits with puri®ed human placental TM as described previously [26]. A 50-fold molar excess amount of the anti-human TM IgG completely inhibited the cofactor activity of puri®ed human native TM for thrombin-dependent protein C activation. Matrigel was purchased from Becton Dickinson (Bedford, MA). Dulbecco's modi®ed Eagle's medium (DMEM) and Dulbecco's phosphate-buffered saline(2) (PBS(2)) were obtained from Nissui Seiyaku Co. (Tokyo, Japan). Fetal calf serum (FCS) was obtained from
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Boehringer Mannheim (Mannheim, Germany). BDF1 mice weighing about 22 g were purchased from Charles River Japan Inc.(Atsugi, Japan). 2.2. Preparation of TMs Active TM puri®ed from human urine (uTM; speci®c activity . 5000 TMU/mg) was donated by Mochida Pharmaceutical Co. Ltd., Tokyo, Japan. One unit of TM (TM unit: TMU) was de®ned as the amount of TM which results in the complete activation of protein C in the assay of thrombin-catalyzed protein C activation as previously described [16]. The puri®ed uTM showed two bands of 72 and 79 kDa on SDS±PAGE under non-reducing conditions and contained the lectin-like domain, the EGF-like domain and part of the O-glycosylation site-rich domain, but lacked the transmembrane and cytoplasmic domains of native TM [16,18]. Recombinant TMs (rTMa and rTMb), which contains the lectin-like domain, the EGF-like domain and the O-glycosylation site-rich domain, but lack the transmembrane and cytoplasmic domains, were prepared by a slightly modi®ed method of Nawa et al. [27]. Brie¯y, rTMs including rTMa (67 kDa without chondroitin-4sulfate glycosaminoglycan on O-glycosylation siterich domain) and rTMb (heterogeneous molecules of 75±95 kDa with chondroitin-4-sulfate glycosaminoglycan on the O-glycosylation site-rich domain) were expressed by transfection of pRS7TM into CHO-K1 Chinese hamster ovary (CHO) cells. The rTMs secreted into cultured medium were separated by chromatography on a Q-Sepharose Fast Flow (Amersham±Pharmacia Biotech Ltd., Buckinghamshire, UK) column equilibrated with 20 mM Tris± HCl (pH 7.4) containing 0.15 M NaCl. rTMa was eluted at 0.35 M NaCl as a sharp peak and rTMb was eluted at 0.6±1.0 M NaCl as a broad peak using a linear gradient from 0.15 to 1.2 M NaCl. Each rTM was further puri®ed by DIP-thrombin Sepharose and Sephadex G-200 (Amersham-Pharmacia Biotech) chromatography. The puri®ed rTMs contained 491 amino acids from NH2-terminal Ala-1 to COOH-terminal Ala-491 [27]. The recombinant EGF-like domain (rTME1±6) of native human TM was prepared as described previously [28]. Brie¯y, rTME1±6 was expressed by transfection of pPG3TM plasmid into CHO-K1 cells, and the secreted rTME1±6 was puri®ed
233
from the culture medium by the methods described for rTMa. rTME1±6 (40 kDa) contains 253 amino acids from NH2-terminal Gln-214 to COOH-terminal Lys466, identical to the known EGF-like domain [28]. The cofactor activities of rTMa, rTMb and rTME1±6 were 5180, 5310 and 4910 TMU/mg protein, respectively. 2.3. Cell culture and cell preparation Murine melanoma B16F10 cells were obtained from American Type Tissue Culture (Rockville, MD). B16F10 cells were grown in DMEM supplemented with 10% FCS, 50 units/ml penicillin and 50 mg/ml streptomycin at 378C in a humidi®ed atmosphere of 5% CO2. The con¯uent B16F10 cells were detached by short exposure to PBS(2) containing 1 mM EDTA and collected by centrifugation. The cells were rinsed twice with 0.1% BSA±DMEM to remove residual EDTA and FCS. Cell suspensions (2 £ 105 /100 ml) in 0.1% BSA±DMEM were used for the invasion assay. 2.4. Cell invasion assay The invasion of B16F10 cells in vitro was measured according to the method reported previously [29,30], with some modi®cation. Cell Culture Inserts (pore size, 8 mm; Becton Dickinson) equipped with a microporous membrane were coated with Matrigel (10 mg/®lter) and dried. The dried Matrigel was reconstituted with 100 ml of FCS-free DMEM containing 0.1% BSA in the absence or presence of tested reagents (various soluble TMs, thrombin, antihuman TM IgG and/or hirudin) for 1 h at room temperature. After removing the solution, 100 ml of 0.1% BSA±DMEM in the absence or presence of tested reagents (two times higher concentrations of the reagents used in reconstitution of Matrigel) and 100 ml of B16F10 cells (2 £ 105 cells) suspended in 0.1% BSA±DMEM were placed in the upper compartments of Cell Culture Insert and then 0.1% BSA±DMEM (650 ml) was added to the lower compartments of Cell Culture Insert. Thus, both the reconstituted Matrigel and the cell suspension in upper compartment contained the same concentration of test reagents. The cells were incubated for 8 h at 378C in 5% CO2. At the end of incubation, the cells on the upper surface of the ®lter were completely
234
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
removed by wiping with a cotton swab. The invaded cells that remained on the lower surface of the ®lter were ®xed with methanol and stained with Diff-Ouick (Green Cross, Osaka, Japan). The stained cells were counted in ®ve randomly selected microscopic ®elds ( £ 200) per ®lter. The experiments were performed in ®ve replicates, and repeated at least three times separately to con®rm reproducibility. 2.5. Measurement of amounts of thrombin and prothrombin Conditioned medium of cultured B16F10 cells (70 ml) or 0.1 mg/ml Matrigel in 0.1% BSA±DMEM (70 ml) were added into 70 ml of 50 mM Tris±HCl (pH 8.3) containing 0.15 M NaCl, 2 mM CaCl2, 0.1% BSA and S-2238 (®nal 300 mM), a chromogenic substrate for thrombin. After incubation for 30 min at 378C, the reaction was stopped by adding 60 ml of 50% acetic acid, followed by measurement of absorbance at 405 nm [31]. In case of thrombin activity assay on cell surface of B16F10 cells, S-2238 (®nal 300 mM) in 50 mM Tris±HCl (pH 8.3) containing 0.15 M NaCl, 2 mM CaCl2, 0.1% BSA was added to the con¯uent cells (0:9 £ 105 cells in a 96-well plate) after three washes with PBS(2). For the prothrombin assay, the conditioned medium, Matrigel and con¯uent cells were treated with ecarin (®nal 0.001 unit/ml) to activate prothrombin to thrombin for 10 min at 378C before addition of S-2238 solution [32]. 2.6. Experimental lung metastasis Experimental lung metastasis was assayed by means of tumor cell injection into the lateral tail vein of mice. Seven BDF1 mice per group were given an intravenous injection of 1 £ 105 B16F10 cells/200 ml admixed with various concentration of uTM or vehicle (PBS(2), pH 7.0). In addition, uTM or vehicle was injected intravenously once a day for 2 days (total three times) after the cell inoculation. Mice were killed under anesthesia at 14 days after the inoculation of the cells. The lungs were removed, washed with PBS(2) and ®xed in 10% neutral formalin at room temperature. The lung metastatic colonies were counted under a dissecting microscope.
2.7. Statistical analysis The statistical signi®cance of differences between the groups was determined by applying Student's ttest.
3. Results 3.1. Effect of uTM on the invasion of B16F10 cells The effect of uTM on the invasion of B16F10 cells through a reconstituted basement membrane (Matrigel) was investigated. The B16F10 cells did not express detectable TM on the cell surface or in cell homogenates ,8:3 £ 1024 TMU or 0.15 ng TM=2 £ 105 cells). This compares with expression levels of 0.22 TMU or 40 ng TM=2 £ 105 cultured human umbilical vein endothelial cells. In this experimental series, the incubation mixture in the upper compartment of Cell Culture Insert contains the same concentration of test reagents, including uTM, antibodies, hirudin and/or rTMs, which were used in reconstitution of the Matrigel. When Matrigel alone was used as a control in the invasion assay of B16F10 cells, approximately 395 ^ 20 cells/®eld invaded to the lower surface of the Matrigel after 8 h. When various concentrations of uTM (10±1000 ng/ml) were added during reconstitution of the Matrigel, the number of cells that crossed the barrier reduced in a dose-dependent manner from 10 to 100 ng/ml of uTM (Fig. 1). Matrigel which contained more than 30 ng/ml uTM signi®cantly reduced the number of invaded B16F10 cells when compared with control Matrigel without uTM. The maximum inhibition of about 50% was observed when Matrigel contained at least 100 ng/ml uTM (Fig. 1). This inhibitory effect of uTM was eliminated by pre-incubation of uTM with rabbit anti-human TM IgG (100 mg/ml), an excess suf®cient to inhibit the cofactor activity of 100 ng/ml uTM (Fig. 2). In control wells, uTM preincubated with pre-immune IgG before reconstitution of Matrigel did not affect the uTM-induced inhibition of cell invasion. These results indicated that the inhibitory action of uTM on the cell invasion was not due to unknown materials in the preparation and was dependent on uTM. Furthermore, hirudin had no effect on the inhibitory action of uTM. The hirudin
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Fig. 1. Effect of uTM on invasion of B16F10 cells into Matrigel. The coated Matrigel (10 mg/®lter) was reconstituted with 100 ml of uTM (0, 10, 30, 100, 300, 1000 ng/ml) in serum-free medium containing 0.1% BSA for 1 h at room temperature. After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of uTM (®nal 0, 10, 30, 100, 300, 1000 ng/ml) were seeded onto ®lters in the upper surface of Cell Culture Insert. After 8 h incubation, the cells invading the lower surface were ®xed and assayed for invasion by using DiffQuick stain as described in Section 2. Experiments were repeated four times separately. Values are indicated as the mean ^ SD. *P , 0:05 compared with control, **P , 0:01 compared with control.
Fig. 2. Effect of hirudin or rabbit anti-human TM IgG on the inhibition of B16F10 cells invasion by uTM. Pre-incubation of uTM (®nal 100 ng/ml) with hirudin (®nal 0.1 unit/ml), rabbit anti-human TM IgG (®nal 100 mg/ml) or rabbit pre-immune IgG (®nal 100 mg/ml) was performed on ice for 30 min, and the solutions were added to the upper compartments 1 h before the cells were plated on the coated Matrigel (10 mg/®lter). After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of pre-treated uTMs (®nal 100 ng/ml) were seeded onto the ®lters. After 8 h incubation, the cells that invaded the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Experiments were repeated three times. Values are indicated as the mean ^ SD. **P , 0:01 compared with control.
235
was present at 0.1 unit/ml, a concentration suf®cient to completely inhibit the proteolytic/amidolytic activity of 0.01 NIH unit/ml thrombin. In control studies, there was no detectable thrombin amidolytic activity ,0:001 NIH unit/ml) nor detectable ecarin-activatable prothrombin in either the conditioned medium of cultured B16F10 cells (0:9 £ 105 cells), the membrane surface of cultured B16F10 cells (0:9 £ 105 cells) or in the Matrigel (0.1 mg/ml) (date not shown). Taken together, these results suggested that uTM repressed the invasion of B16F10 cells through Matrigel by thrombin-independent mechanism. To determine the structural requirements for this inhibitory action of uTM, the effect of various recombinant human soluble TMs on the invasion of B16F10 cells was compared with that of uTM (Fig. 3). Three recombinant soluble TMs, rTMa (67 kDa), rTMb (75±95 kDa), and rTME1±6 (40 kDa), were used in this study. Both rTMa and rTMb consist of the lectin-like, EGF-like and O-glycosylation site-rich domains of native TM, although rTMb has the chondroitin-4-sulfate glycosaminoglycan bound the O-glycosylation site-rich domain. rTME1±6 represents the EGF-like domain of native human TM. As shown in Fig. 3, Matrigel reconstituted with 1.3 nM rTMa or
Fig. 3. Effects of uTM and rTMs on invasion of B16F10 cells into Matrigel. The coated Matrigel (10 mg/®lter) was reconstituted with 100 ml of 1.3 nM soluble TMs (uTM, rTMa, rTMb and rTME1±6) in serum free medium containing 0.1% BSA for 1 h at room temperature. After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of each soluble TM (®nal 1.3 nM) were seeded onto the ®lters. After 8 h incubation, the cells on the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Each column represents the mean ^ SD of four values. *P , 0:05 compared with control.
236
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
rTMb signi®cantly decreased the invasion cells to 65± 70%, similar the Matrigel containing 1.3 nM uTM (100 ng/ml). In contrast, Matrigel containing 1.3 nM rTME1±6 did not inhibit the cell invasion. 3.2. Effect of uTM on the invasion of B16F10 cells in the presence of thrombin An earlier report demonstrated that high concentrations of thrombin accelerate adhesion of B16F10 cells to ®bronectin and metastasis of the cells [33]. Thus, the effects of uTM and/or hirudin on the invasion of B16F10 cells in Matrigel containing a high concentration of thrombin were investigated. Addition of thrombin (1 NIH unit/ml, approx. 10 nM) into Matrigel signi®cantly accelerated the invasion of B16F10 cells 1.5-fold when compared with the cell invasion level with the control Matrigel (Fig. 4). The thrombinenhanced invasion of B16F10 cells was decreased by addition of uTM (100 or 1000 ng/ml, approx. 1.3 or 13 nM) to the thrombin-containing Matrigel by 50% and 100%, respectively, relative to that observed with thrombin alone (Fig. 4). Addition of 10 units/ml
hirudin into the thrombin-contained Matrigel also completely abolished the accelerating action of thrombin on B16F10 cell invasion (Fig. 4). Moreover, co-addition of uTM (100 or 1000 ng/ml) with hirudin (10 units/ml) into the thrombin-containing Matrigel additionally enhanced the inhibitory activity (Fig. 4). These results indicated that while the thrombin-enhanced invasion activity was dependent on the proteolytic activity of thrombin, the inhibitory action of uTM was also exhibited in absence of thrombin proteolytic activity. 3.3. Effect of uTM on experimental lung metastasis of B16F10 cells The ability of uTM to inhibit experimental lung metastasis after intravenous injection of B16F10 cells in mice was investigated (Fig. 5). In this model, 11.6 ^ 1.9 colonies were observed per lung in control mice at 14 days after injection with B16F10 cells. In the case of single co-injection of uTM (1.74 mg uTM/kg of body weight) with B16F10 cells, the metastatic colonies of B16F10 cells in the lung was suppressed to 6.4 ^ 3.6 colonies/lung, but the reduction was not statistically signi®cant. After co-injection of 1.74 or 5.21 mg uTM/kg with B16F10 cells, additional injection of same amount of uTM, once a day up to 2 days (total three times), signi®cantly inhibited colony formation of B16F10 cells in the lung to 42 and 35%, respectively, of the colonies found in the control mice. 4. Discussion
Fig. 4. Effect of uTM on thrombin-stimulated B16F10 cell invasion. Pre-incubation of uTM (®nal 100, 1000 ng/ml) or hirudin (®nal 10 units/ml) with thrombin (®nal 1 NIH unit/ml) was performed for 30 min at 378C, and the solutions were added to the upper compartments 1 h before the cells were plated on the coated Matrigel (10 mg/®lter). After removing the solution, B16F10 cells (2 £ 105 cells/ 100 ml) and 100 ml of uTM (®nal 100, 1000 ng/ml) or hirudin (®nal 10 units/ml) pre-incubated with thrombin (®nal 1 NIH unit/ml) were seeded onto the ®lters. After 8 h, the cells invading the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Each column represents the mean ^ SD of four values. *P , 0:05 compared with control, **P , 0:01 compared with control, ²P , 0:01 compared with thrombin alone.
The present study indicated that Matrigel containing exogenous uTM inhibits invasion of B16F10 cells into the reconstituted basement membrane in vitro (Fig. 1). It is known that TM inhibits the procoagulant activity of thrombin and acts as a cofactor of thrombin-catalyzed protein C and TAFI activations and thrombin-catalyzed inactivation of single chain uPA. Therefore, we investigated whether the inhibitory action of uTM on tumor cell invasion was dependent on thrombin. No detectable activity of prothrombin/ thrombin was observed in B16F10 cells and Matrigel, and the inhibitory effect of uTM on the cell invasion was not interfered even in the presence of hirudin, a thrombin inhibitor (Fig. 2). Additionally, the throm-
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Fig. 5. Effect of uTM on experimental lung metastases in mice. (A) B16F10 cells (1 £ 105 cells) were injected with or without uTM (0, 0.521, 1.74, or 5.21 mg/kg) into the lateral tail vein of BDF1 mice. Various dose of uTM were injected i.v. once a day for 2 days. A single administration of uTM (1.74 mg/kg) was coinjected with the cells i.v.. After 14 days, the mice (n 7) were sacri®ced and the lungs were removed. The lungs were ®xed and observed for metastasis as described in Section 2. Values are indicated as the mean ^ SD. *P , 0:05 compared with control, **P , 0:01 compared with control. (B,C) Representative photographs of mouse lung treated with or without uTM. (B) Control; (C) mice treated with 3 injections of uTM (5.21 mg/kg). Black
bin-enhanced invasion of B16F10 cells was also reduced by co-existence of uTM and hirudin in Matrigel to less than the invasion level observed in the control Matrigel (Fig. 4). These results suggest that the inhibitory action of uTM on cell invasion of B16F10 cells is due to a thrombin-independent mechanism. Tumor cell invasion consists of at least three steps,
237
including attachment of tumor cells to matrix, local degradation of matrix by tumor associated proteases and locomotion of tumor cells into tissues [1±6]. Rabhi-Sabile et al. [34] has reported that the chondroitin-4-sulfate of TM is an important moiety in cytoadhesion between Plasmodium falciparum-infected erythrocytes and TM-expressed cells, suggesting that the chondroitin-4-sulfate of TM plays an important role in interaction with cells. In addition, it has been noted that large chondroitin sulfate proteoglycans inhibited cell-matrix interaction [35,36]. Therefore, we examined the effect of chondroitin-4-sulfate of TM on B16F10 cell invasion by comparison of the inhibitory effects of rTMa and rTMb. The present data showed that the intensity of inhibitory effect of rTMb containing chondroitin-4-sulfate were similar to that of rTMa lacking chondroitin-4-sulfate (Fig. 3). Therefore, the inhibitory action of the soluble TMs on B16F10 cell invasion did not depend on the chondroitin-4-sulfate glycosaminoglycan attached to the O-glycosylation site-rich domain of the soluble TMs. Imada et al. [37] has noted that fetomodulin, which is identical to TM, appeared during differentiation of mouse teratocarcinoma F9 cells (an embryonal carcinoma stem cell line) into parietal endoderm cells, and suggested that fetomodulin/TM expressed on the surface of differentiated cells may act as a multifunctional protein during cell±cell interactions. It has further been postulated that the lectin-like domain of TM expressed on tumor cell surfaces could interact with cellular membranes and an extracellular matrix protein and may act as an adhesion molecule [22,24]. Interestingly, rTME1±6 lacking lectin-like domain did not inhibit B16F10 cell invasion (Fig. 3). The result implied that the lectin-like domain of soluble TM might contribute to interfere with attachment and/or locomotion of B16F10 cells to matrix through the interaction with some molecule in Matrigel or on the cell surface in the invasion process. However, we could not prove the mechanism in detail, because we can not yet examine the effect of lectin-like domain of soluble TM on B16F10 cell invasion. Additional studies are necessary to evaluate whether the lectin-like domain participates in the inhibitory action of soluble TM on tumor cell invasion. Previous papers reported that several inhibitors against Matrigel invasion of murine melanoma,
238
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
including B16F10 cells, also suppressed experimental lung metastasis in mice [36,38]. Therefore, it was possible that uTM may have similar effects in this murine model. Our experimental metastasis model in this study was mild compared with that in previous reports with respect to murine melanoma [36,38], because the metastatic colonies in control mice at 14 days after injection of B16F10 cells were about 12 colonies/lung. In this model, a single co-injection of 1.74 mg uTM/kg body weight with B16F10 cells showed a tendency to suppress the lung metastasis of the cells without statistical signi®cance, while additional injections of uTM, once a day for up to two days, more strongly and signi®cantly inhibited the lung metastasis (Fig. 5). It has been reported that micrometastasis is observed at several days postinjection of tumor cells in experimental lung metastasis model [1]. Therefore, the repeat injections of uTM for signi®cant inhibition of the lung metastasis may be required to maintain adequate plasma concentration of uTM in the invasion process of the tumor cells from plasma to extracellular matrix, although we do not yet know the half-life of uTM in the plasma of the mice injected with uTM. Most of tumor cells are capable of generating thrombin through activation of plasma coagulation cascade by tissue factor and/or tumor-associate proteases expressed on cell surface [39]. Esumi et al. [40] have demonstrated that injection of recombinant hirudin, a thrombin inhibitor, signi®cantly inhibited experimental lung metastasis of B16F10 cells, and concluded that the inhibition was related with inhibition of coagulation events by the inhibitor. In addition, the metastatic potential of B16F10 cells were modulated by uPA secreted from the cells [41,42]. Although the present in vitro experiments suggested that uTM inhibited Matrigel invasion of B16F10 cells in a thrombin-independent mechanism, we can not exclude a possibility that the inhibitory action of injected uTM on experimental lung metastasis of B16F10 cells in vivo may be resulted from inhibition of coagulation and/or ®brinolysis events in circulating plasma through the thrombin-dependent mechanism other than the thrombin-independent mechanism, since circulating plasma contains various proenzymes and active enzymes such as protein C, TAFI and uPA which are activated or inactivated by thrombin±TM complex.
The present study indicated that Matrigel containing uTM, puri®ed from human urine, inhibited the invasion of B16F10 cells in vitro by a thrombin-independent mechanism and intravenous injection of uTM suppressed tumor metastasis in an experimental metastasis model in mice. It remains the possibility that TM expressed on tumor cell surfaces also may regulate the invasion and metastasis of tumor cells. Experiments are now in progress to elucidate the detailed mechanism of suppression of invasion and metastasis of tumor cells by TM. Acknowledgements This work was partly supported by Grant-in-Aid for Encouragement of Young Scientists from The Ministry of Education, Science, Sports and Culture Japan Society for the Promotion of Science (grant nos. 09771997 and 11771450) and by the Sasagawa Scienti®c Research Grant from The Japan Science Society (grant no. 11-180). References [1] I.J. Fidler, Tumor heterogeneity and the biology of cancer invasion and metastasis, Cancer Res. 38 (1978) 2651±2660. [2] L.A. Liotta, Tumor invasion and metastasis: role of the extracellular matrix: Rhoads memorial award lecture, Cancer Res. 46 (1986) 1±7. [3] V.P. Terranova, E.S. Hujanen, G.R. Martin, Basement membrane and the invasive activity of metastatic tumor cells, J. Natl. Cancer Inst. 77 (1986) 311±316. [4] E. Bastida, A. Ordinas, Platelet contribution to the formation of metastatic foci: the role of cancer cell-induced platelet activation, Haemostasis 18 (1988) 29±36. [5] R.H. Kramer, G.L. Nicolson, Interaction of tumor cells with vascular endothelial cell monolayers: a model for metastatic invasion, Proc. Natl. Acad. Sci. USA 76 (1979) 5704±5708. [6] D.G. Menter, B.W. Steinert, B.F. Sloane, N. Gundlach, C.Y. O'Gara, L.J. Marnett, C. Diglio, D. Walz, J.D. Taylor, K.V. Honn, Role of platelet membrane in enhancement of tumor cell adhesion to endothelial cell extracellular matrix, Cancer Res. 47 (1987) 6751±6762. [7] C.T. Esmon, The roles of protein C and thrombomodulin in the regulation of blood coagulation, J. Biol. Chem. 264 (1989) 4743±4746. [8] K. Suzuki, H. Kusumoto, Y. Deyashiki, J. Nishioka, I. Maruyama, M. Zushi, S. Kawahara, G. Honda, S. Yamamoto, S. Horiguchi, Structure and expression of human thrombomodulin, a thrombin receptor on endothelium acting as a cofactor for protein C activation, EMBO J. 6 (1987) 1891±1897.
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240 [9] D. Wen, W.A. Dittman, R.D. Ye, L.L. Deaven, P.W. Majerus, J.E. Sadler, Human thrombomodulin: complete cDNA sequence and chromosome localization of the gene, Biochemistry 26 (1987) 4350±4357. [10] M. Zushi, K. Gomi, S. Yamamoto, I. Maruyama, T. Hayashi, K. Suzuki, The last three consecutive epidermal growth factor-like structures of human thrombomodulin comprise the minimum functional domain for protein C-activating cofactor activity and anticoagulant activity, J. Biol. Chem. 264 (1989) 10351±10353. [11] M. Tsiang, S.R. Lentz, J.E. Sadler, Functional domains of membrane-bound human thrombomodulin. EGF-like domains four to six and the serine/threonine-rich domain are required for cofactor activity, J. Biol. Chem. 267 (1992) 6164±6170. [12] H. Ishii, P.W. Majerus, Thrombomodulin is present in human plasma and urine, J. Clin. Invest. 76 (1985) 2178±2181. [13] H. Ishii, H. Uchiyama, M. Kazama, Soluble thrombomodulin antigen in conditioned medium is increased by damage of endothelial cells, Thromb. Haemost. 65 (1991) 618±623. [14] M. Nakano, M. Furutani, S. Hiraishi, H. Ishii, Characterization of soluble thrombomodulin fragments in human urine, Thromb. Haemost. 79 (1998) 331±337. [15] D.E. Jackson, T.J. Teras, H.H. Salem, C.A. Mitchel, Puri®cation and characterization of two form of soluble TM from human urine, Eur. J. Biochem. 221 (1994) 1079±1087. [16] Y. Takahashi, Y. Hosaka, H. Niina, K. Nagasawa, M. Naotsuka, K. Sakai, A. Uemura, Soluble thrombomodulin puri®ed from human urine exhibits a potent anticoagulant effect in vitro and in vivo, Thromb. Haemost. 73 (1995) 805±811. [17] Y. Takahashi, Y. Hosaka, K. Imada, T. Adachi, H. Niina, M. Watanabe, H. Mochizuki, Human urinary soluble thrombomodulin (MR-33) improves disseminated intravascular coagulation without affecting bleeding time in rats: comparison with low molecular weight heparin, Thromb. Haemost. 77 (1997) 789±795. [18] H. Wakabayashi, S. Natsuka, T. Mega, N. Otsuki, M. Isaji, M. Naotsuka, S. Koyama, T. Kanamori, K. Sakai, S. Hase, Novel proteoglycan linkage tetrasaccharides of human urinary soluble thrombomodulin, SO4-3GlcAb1±3Galb1-3(1/2Siaa2± 6)Galb1±4Xyl, J. Biol. Chem. 279 (1999) 5436±5442. [19] G.A.W. de Munk, E. Groeneveld, D.C. Rijken, Acceleration of the thrombin inactivation of single chain urokinase-type plasminogen activator (pro-urokinase) by thrombomodulin, J. Clin. Invest. 88 (1991) 1680±1684. [20] L. Bajzar, J. Morser, M.E. Nesheim, TAFI, or plasma procarboxypeptidase B, couples the coagulation and ®brinolytic cascades through the thrombin-thrombomodulin complex, J. Biol. Chem. 271 (1996) 16603±16608. [21] Y. Hosaka, Y. Takahashi, H. Ishii, Thrombomodulin in human plasma contributes to inhibit ®brinolysis through acceleration of thrombin-dependent activation of plasma procarboxypeptidase B, Thromb. Haemost. 79 (1998) 371±377. [22] T. Suehiro, M. Shimada, T. Matsumata, A. Taketomi, K. Yamamoto, K. Sugimachi, Thrombomodulin inhibits intrahepatic spread in human hepatocellular carcinoma, Hepatology 21 (1995) 1285±1290. [23] S. Wilhelm, M. Schmitt, J. Parkinson, W. Kuhn, H. Graeff,
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31] [32] [33]
[34]
[35]
239
O.G. Wilhelm, Thrombomodulin, a receptor for the serine protease thrombin, is decreased in primary tumors and metastases but increased in ascitic ¯uids of patients with advanced ovarian cancer FIGO IIIc, Int. J. Oncol. 13 (1998) 645± 651. M. Tabata, K. Sugihara, S. Tonezawa, S. Yamashita, I. Maruyama, An immunohistochemical study of thrombomodulin in oral squamous cell carcinoma and its association with invasive and metastatic potential, J. Oral. Pathol. Med. 26 (1997) 258±264. S.J. Kim, E. Shiba, H. Ishii, T. Inoue, T. Taguchi, Y. Tanji, Y. Kimoto, M. Izukura, S. Takai, Thrombomodulin is a new biological and prognostic marker for breast cancer: an immunohistochemical study, Anticancer Res. 17 (1997) 2319± 2323. H. Ishii, M. Nakano, J. Tsubouchi, T. Ishikawa, H. Uchiyama, S. Hiraishi, C. Tahara, Y. Miyajima, M. Kazama, Establishment of enzyme immunoassay of human thrombomodulin in plasma and urine using monoclonal antibodies, Thromb. Haemost. 63 (1990) 157±162. K. Nawa, K. Sakano, H. Fujiwara, Y. Sato, N. Sugiyama, T. Teruuchi, M. Iwamoto, Y. Marumoto, Presence and function of chondroitin-4-sulfate on recombinant human soluble thrombomodulin, Biochem. Biophys. Res. Commun. 171 (1990) 729±737. H. Hamada, H. Ishii, K. Sakyo, S. Horie, K. Nishiki, M. Kazama, The epidermal growth factor-like domain of recombinant human thrombomodulin exhibits mitogenic activity for swiss 3T3 cells, Blood 86 (1995) 225±233. A. Albini, H. Iwamoto, H.K. Kleinman, G.R. Martin, S.A. Aaronson, J.M. Kozolowski, R.N. McEwan, A rapid in vitro assay for quantitating the invasive potential of tumor cells, Cancer Res. 47 (1987) 3239±3245. L.M. Bafetti, T.N. Young, Y. Itoh, M.S. Stack, Intact vitronectin induces matrix metalloproteinase-2 and tissue inhibitor of metalloproteinases-2 expression and enhanced cellular invasion by melanoma cells, J. Biol. Chem. 273 (1998) 143±149. Q. Wu, M. Tsiang, S.R. Lentz, J.E. Sadler, Ligand speci®city of human thrombomodulin, J. Biol. Chem. 267 (1992) 7083± 7088. T. Morita, S. Iwanaga, Prothrombin activator from Echis carinatus venom, Methods Enzymol. 80 (1981) 303±311. M.L. Nierodzik, K. Chen, K. Takeshita, J.J. Li, Y.Q. Huang, X.S. Feng, M.R. D'Andrea, P. Andrade-Gordon, S. Karpatkin, Protease-activated receptor 1 (PAR-1) is required and ratelimiting for thrombin-enhanced experimental pulmonary metastasis, Blood 92 (1998) 3694±3700. S. Rabhi-Sabile, M. Steiner-Mosonyi, S. Pollefeyt, D. Collen, B. Pouvelle, J. Gysin, M.C. Boffa, E.M. Conway, Plasmodium falciparum-infected erythrocytes: a mutational analysis of cytoadherence via murine thrombomodulin, Thromb. Haemost. 81 (1999) 815±821. N. Sugiura, K. Sakurai, Y. Hori, S. Karasawa, S. Suzuki, K. Kimura, Preparation of lipid-derivatized glycosaminoglycans to probe a regulatory function of the carbohydrate moieties of proteoglycans in cell±matrix interaction, J. Biol. Chem. 268 (1993) 15779±15787.
240
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
[36] K. Karasawa, N. Sugiura, Y. Hori, S. Suzuki, J. Onaya, K. Sakurai, K. Kimata, Inhibition of experimental metastasis and cell adhesion of murine melanoma cells by chondroitin sulfate-derivatized lipid, a neoproteoglycan with anti-cell adhesion activity, Clin. Exp. Metastasis 15 (1997) 83±93. [37] S. Imada, H. Yamaguchi, M. Nagumo, S. Katayanagi, H. Iwasaki, M. Imada, Identi®cation of fetomodulin, a surface marker protein of fetal development, as thrombomodulin by gene cloning and function assays, Dev. Biol. 140 (1990) 113± 122. [38] T. Mishima, J. Murata, M. Toyoshima, H. Fujii, M. Nakajima, T. Hayashi, T. Kato, I. Saiki, Inhibition of tumor invasion and metastasis by calcium spirulan (Ca-SP), a novel sulfated polysaccharide derived from a blue-green alga, Spirulina platensis, Clin. Exp. Metastasis 16 (1998) 541±550.
[39] L.R. Zacharski, V. Costantin, Blood coagulation activation in cancer: challenges for cancer treatment, HaÈemostaseologie 15 (1995) 14±20. [40] N. Esumi, D. Fan, I.J. Fidler, Inhibition of murine melanoma experimental metastasis by recombinant desulfatohirudin, a highly speci®c thrombin inhibitor, Cancer Res. 51 (1991) 4549±4556. [41] V.J. Hearing, L.W. Law, A. Corti, E. Appella, F. Blasi, Modulation of metastatic potential by cell surface urokinase of murine melanoma cells, Cancer Res. 48 (1988) 1270± 1278. [42] H. Yu, R.M. Schultz, Relationship between secreted urokinase plasminogen activator activity and metastatic potential in murine B16 cells transfected with human urokinase sense and antisense genes, Cancer Res. 50 (1990) 7623±7633.
www.elsevier.com/locate/canlet
Inhibition of invasion and experimental metastasis of murine melanoma cells by human soluble thrombomodulin Yoshitaka Hosaka, Toshiyuki Higuchi, Michiko Tsumagari, Hidemi Ishii* Department of Public Health, Showa Pharmaceutical University, Higashi Tamagawa Gakuen, Machida, Tokyo 194-8543, Japan Received 24 July 2000; received in revised form 21 September 2000; accepted 22 September 2000
Abstract Thrombomodulin (TM) is an anticoagulant molecule expressed on the endothelial cell surface and soluble TM antigen, which is present in human plasma and urine, represents the products of limited proteolytic cleavage of cell-surface TM. Recently, it was demonstrated that TM is also expressed on the surface of several tumor cells and the expression level of TM negatively correlated with malignancy in cancer. We investigated the effect of soluble TM isolated from human urine (uTM) on the invasion and metastasis of murine melanoma cells (B16F10 cells) through a reconstituted basement membrane (Matrigel) and in a murine model of experimental lung metastasis. Matrigel reconstituted with uTM inhibited the invasion of B16F10 cells in a dose-dependent manner in a range from 10 to 1000 ng/ml uTM as compared with the control Matrigel without uTM. The inhibitory action of uTM was not altered in the presence of an excess amount of hirudin, an inhibitor of thrombin proteolytic activity, but abolished in the presence of anti-human TM IgG. Matrigel reconstituted with thrombin (1 NIH unit/ml) enhanced the invasion level of cells by 1.5-fold relative to the control Matrigel without thrombin. The thrombin-enhanced invasion of B16F10 cells was repressed by addition of hirudin (10 units/ml) or uTM (100 ng/ml) into the Matrigel. Matrigel reconstituted with hirudin (10 units/ml) and uTM (100 ng/ml) additionally accelerated the inhibitory activity of hirudin or uTM on the thrombin-enhanced invasion of B16F10 cells. Moreover, metastatic colonies formed in the lungs of mice injected intravenously with B16F10 cells were signi®cantly reduced by injection of uTM once a day up to 2 days after co-injection of uTM with the cells. These results suggested that Matrigel reconstituted with uTM inhibited the invasion of B16F10 cells in vitro through a thrombin-independent mechanism and the injection of uTM suppressed experimental lung metastasis of the cells in mice. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Thrombomodulin; Thrombin; Invasion; Metastasis
1. Introduction Tumor metastasis is a complex series of multi-step processes [1] of which invasion of tumor cells through the basement membrane is a critical step in the formation of metastases [2,3]. Tumor invasion consists of at * Corresponding author. Tel.: 181-42-721-1561; fax: 181-42721-1561. E-mail address: [email protected] (H. Ishii).
least three steps, which include attachment of tumor cells to matrix components, local degradation of matrix by tumor associated proteases such as urokinase-type plasminogen activator (uPA) and matrix metalloproteinases (MMPs), and locomotion of tumor cells into the region of the matrix degraded by pericellular proteolysis. Furthermore, in the case of blood-borne metastasis, thrombotic events play an important role in the pathogenesis of metastasis, since the formation of tumor emboli which arises from acti-
0304-3835/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(00)00617-0
232
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
vation of the clotting cascade and platelet aggregation facilitates the attachment and invasion of tumor cells through endothelial cells and the subendothelial matrix [4±6]. Thrombomodulin (TM) is a thrombin receptor found on the surface of vascular endothelial cells. Forming a high-af®nity complex with thrombin, TM directly inhibits the procoagulant activity of thrombin such as ®brin formation and accelerates the thrombindependent activation of the anticoagulant protein C zymogen [7]. Activated protein C inactivates coagulation factor Va and VIIIa in the presence of cofactor protein S which results in reduced thrombin formation [7]. Therefore, it has been postulated that TM is an important natural regulator in maintaining the ¯uidity of circulating blood [7]. Human TM consists of ®ve domains: the lectin-like domain (amino-terminal domain), epidermal growth factor (EGF)-like domain, O-glycosylation site-rich domain and transmembrane and cytoplasmic domains [8,9]. The EGF-like domain is composed of six tandem EGF-like structures having homology with the domain-III of the human EGF precursor [8,9]. The fourth, ®fth and sixth EGF-structures and the chondroitin-4-sulfate glycosaminoglycan bound O-glycosylation site-rich domain participate in the cofactor activity of TM for thrombin-catalyzed activation of protein C [10,11]. TM antigen is found not only on the endothelial cell surface, but also in human plasma and urine as soluble forms, which probably represent the products of limited proteolytic cleavage of cell-surface TM [12± 15]. We isolated the major active form of naturally existing soluble TM from human urine (uTM) which migrated as two bands of 72 and 79 kDa contained the lectin-like domain, the EGF-like domain and part of the O-glycosylation site-rich domain, but lacked the transmembrane and cytoplasmic domains of native TM [16±18]. The inhibitory activity of uTM for procoagulant activity of thrombin and the cofactor activity for thrombin-catalyzed protein C activation are retained. Therefore, uTM exhibits strong anticoagulant activities in vitro and in vivo [16,17]. TM also inhibits ®brinolysis through the acceleration of thrombin-catalyzed inactivation of single chain uPA (latent form of uPA) [19] in vitro and the activation of the thrombin-activatable ®brinolysis inhibitor (TAFI) in a puri®ed system [20] and in plasma [21]. Therefore, TM participates in the regulation of ®brinolysis and
the prevention of tissue damage under conditions where coagulation and/or ®brinolysis is accelerating. Recently, it has been reported that TM is present in several tumors, such as hepatocellular carcinoma [22], ovarian cancer [23] and squamous cell carcinoma [24], and that there is a negative correlation between the expression level of TM and malignancy of the cancer. We also suggested that TM may serve as a prognostic factor for metastatic breast cancer, since patients with low TM expression in breast cancer tissues had a signi®cantly lower rate of disease-free survival than patients with high TM expression [25]. Therefore, it was suggested that TM may play a role as a regulator during tumor invasion and metastasis. However, there is no experimental evidence that TM has inhibitory action against tumor invasion and metastasis. In the current study, we investigated the effects of uTM, a major active form isolated from naturally existing soluble TM antigen in human urine, on invasion and metastasis of murine melanoma cells (B16F10 cells) using reconstituted basement membrane (Matrigel) and an experimental lung metastasis model in mice.
2. Materials and methods 2.1. Materials Reagents were purchased from Wako Pure Chemical Industries (Osaka, Japan), except where indicated. Thrombin, hirudin, ecarin (Echis carinatus venom) and bovine serum albumin (BSA) were obtained from Sigma (St. Louis, MO). Chromogenic substrate S-2238 was obtained from Daiichi Chemical Co. (Tokyo, Japan). Prothrombin was from Enzyme Research Laboratories (South Bend, IN). Polyclonal rabbit anti-human TM IgG was produced by immunization of rabbits with puri®ed human placental TM as described previously [26]. A 50-fold molar excess amount of the anti-human TM IgG completely inhibited the cofactor activity of puri®ed human native TM for thrombin-dependent protein C activation. Matrigel was purchased from Becton Dickinson (Bedford, MA). Dulbecco's modi®ed Eagle's medium (DMEM) and Dulbecco's phosphate-buffered saline(2) (PBS(2)) were obtained from Nissui Seiyaku Co. (Tokyo, Japan). Fetal calf serum (FCS) was obtained from
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Boehringer Mannheim (Mannheim, Germany). BDF1 mice weighing about 22 g were purchased from Charles River Japan Inc.(Atsugi, Japan). 2.2. Preparation of TMs Active TM puri®ed from human urine (uTM; speci®c activity . 5000 TMU/mg) was donated by Mochida Pharmaceutical Co. Ltd., Tokyo, Japan. One unit of TM (TM unit: TMU) was de®ned as the amount of TM which results in the complete activation of protein C in the assay of thrombin-catalyzed protein C activation as previously described [16]. The puri®ed uTM showed two bands of 72 and 79 kDa on SDS±PAGE under non-reducing conditions and contained the lectin-like domain, the EGF-like domain and part of the O-glycosylation site-rich domain, but lacked the transmembrane and cytoplasmic domains of native TM [16,18]. Recombinant TMs (rTMa and rTMb), which contains the lectin-like domain, the EGF-like domain and the O-glycosylation site-rich domain, but lack the transmembrane and cytoplasmic domains, were prepared by a slightly modi®ed method of Nawa et al. [27]. Brie¯y, rTMs including rTMa (67 kDa without chondroitin-4sulfate glycosaminoglycan on O-glycosylation siterich domain) and rTMb (heterogeneous molecules of 75±95 kDa with chondroitin-4-sulfate glycosaminoglycan on the O-glycosylation site-rich domain) were expressed by transfection of pRS7TM into CHO-K1 Chinese hamster ovary (CHO) cells. The rTMs secreted into cultured medium were separated by chromatography on a Q-Sepharose Fast Flow (Amersham±Pharmacia Biotech Ltd., Buckinghamshire, UK) column equilibrated with 20 mM Tris± HCl (pH 7.4) containing 0.15 M NaCl. rTMa was eluted at 0.35 M NaCl as a sharp peak and rTMb was eluted at 0.6±1.0 M NaCl as a broad peak using a linear gradient from 0.15 to 1.2 M NaCl. Each rTM was further puri®ed by DIP-thrombin Sepharose and Sephadex G-200 (Amersham-Pharmacia Biotech) chromatography. The puri®ed rTMs contained 491 amino acids from NH2-terminal Ala-1 to COOH-terminal Ala-491 [27]. The recombinant EGF-like domain (rTME1±6) of native human TM was prepared as described previously [28]. Brie¯y, rTME1±6 was expressed by transfection of pPG3TM plasmid into CHO-K1 cells, and the secreted rTME1±6 was puri®ed
233
from the culture medium by the methods described for rTMa. rTME1±6 (40 kDa) contains 253 amino acids from NH2-terminal Gln-214 to COOH-terminal Lys466, identical to the known EGF-like domain [28]. The cofactor activities of rTMa, rTMb and rTME1±6 were 5180, 5310 and 4910 TMU/mg protein, respectively. 2.3. Cell culture and cell preparation Murine melanoma B16F10 cells were obtained from American Type Tissue Culture (Rockville, MD). B16F10 cells were grown in DMEM supplemented with 10% FCS, 50 units/ml penicillin and 50 mg/ml streptomycin at 378C in a humidi®ed atmosphere of 5% CO2. The con¯uent B16F10 cells were detached by short exposure to PBS(2) containing 1 mM EDTA and collected by centrifugation. The cells were rinsed twice with 0.1% BSA±DMEM to remove residual EDTA and FCS. Cell suspensions (2 £ 105 /100 ml) in 0.1% BSA±DMEM were used for the invasion assay. 2.4. Cell invasion assay The invasion of B16F10 cells in vitro was measured according to the method reported previously [29,30], with some modi®cation. Cell Culture Inserts (pore size, 8 mm; Becton Dickinson) equipped with a microporous membrane were coated with Matrigel (10 mg/®lter) and dried. The dried Matrigel was reconstituted with 100 ml of FCS-free DMEM containing 0.1% BSA in the absence or presence of tested reagents (various soluble TMs, thrombin, antihuman TM IgG and/or hirudin) for 1 h at room temperature. After removing the solution, 100 ml of 0.1% BSA±DMEM in the absence or presence of tested reagents (two times higher concentrations of the reagents used in reconstitution of Matrigel) and 100 ml of B16F10 cells (2 £ 105 cells) suspended in 0.1% BSA±DMEM were placed in the upper compartments of Cell Culture Insert and then 0.1% BSA±DMEM (650 ml) was added to the lower compartments of Cell Culture Insert. Thus, both the reconstituted Matrigel and the cell suspension in upper compartment contained the same concentration of test reagents. The cells were incubated for 8 h at 378C in 5% CO2. At the end of incubation, the cells on the upper surface of the ®lter were completely
234
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
removed by wiping with a cotton swab. The invaded cells that remained on the lower surface of the ®lter were ®xed with methanol and stained with Diff-Ouick (Green Cross, Osaka, Japan). The stained cells were counted in ®ve randomly selected microscopic ®elds ( £ 200) per ®lter. The experiments were performed in ®ve replicates, and repeated at least three times separately to con®rm reproducibility. 2.5. Measurement of amounts of thrombin and prothrombin Conditioned medium of cultured B16F10 cells (70 ml) or 0.1 mg/ml Matrigel in 0.1% BSA±DMEM (70 ml) were added into 70 ml of 50 mM Tris±HCl (pH 8.3) containing 0.15 M NaCl, 2 mM CaCl2, 0.1% BSA and S-2238 (®nal 300 mM), a chromogenic substrate for thrombin. After incubation for 30 min at 378C, the reaction was stopped by adding 60 ml of 50% acetic acid, followed by measurement of absorbance at 405 nm [31]. In case of thrombin activity assay on cell surface of B16F10 cells, S-2238 (®nal 300 mM) in 50 mM Tris±HCl (pH 8.3) containing 0.15 M NaCl, 2 mM CaCl2, 0.1% BSA was added to the con¯uent cells (0:9 £ 105 cells in a 96-well plate) after three washes with PBS(2). For the prothrombin assay, the conditioned medium, Matrigel and con¯uent cells were treated with ecarin (®nal 0.001 unit/ml) to activate prothrombin to thrombin for 10 min at 378C before addition of S-2238 solution [32]. 2.6. Experimental lung metastasis Experimental lung metastasis was assayed by means of tumor cell injection into the lateral tail vein of mice. Seven BDF1 mice per group were given an intravenous injection of 1 £ 105 B16F10 cells/200 ml admixed with various concentration of uTM or vehicle (PBS(2), pH 7.0). In addition, uTM or vehicle was injected intravenously once a day for 2 days (total three times) after the cell inoculation. Mice were killed under anesthesia at 14 days after the inoculation of the cells. The lungs were removed, washed with PBS(2) and ®xed in 10% neutral formalin at room temperature. The lung metastatic colonies were counted under a dissecting microscope.
2.7. Statistical analysis The statistical signi®cance of differences between the groups was determined by applying Student's ttest.
3. Results 3.1. Effect of uTM on the invasion of B16F10 cells The effect of uTM on the invasion of B16F10 cells through a reconstituted basement membrane (Matrigel) was investigated. The B16F10 cells did not express detectable TM on the cell surface or in cell homogenates ,8:3 £ 1024 TMU or 0.15 ng TM=2 £ 105 cells). This compares with expression levels of 0.22 TMU or 40 ng TM=2 £ 105 cultured human umbilical vein endothelial cells. In this experimental series, the incubation mixture in the upper compartment of Cell Culture Insert contains the same concentration of test reagents, including uTM, antibodies, hirudin and/or rTMs, which were used in reconstitution of the Matrigel. When Matrigel alone was used as a control in the invasion assay of B16F10 cells, approximately 395 ^ 20 cells/®eld invaded to the lower surface of the Matrigel after 8 h. When various concentrations of uTM (10±1000 ng/ml) were added during reconstitution of the Matrigel, the number of cells that crossed the barrier reduced in a dose-dependent manner from 10 to 100 ng/ml of uTM (Fig. 1). Matrigel which contained more than 30 ng/ml uTM signi®cantly reduced the number of invaded B16F10 cells when compared with control Matrigel without uTM. The maximum inhibition of about 50% was observed when Matrigel contained at least 100 ng/ml uTM (Fig. 1). This inhibitory effect of uTM was eliminated by pre-incubation of uTM with rabbit anti-human TM IgG (100 mg/ml), an excess suf®cient to inhibit the cofactor activity of 100 ng/ml uTM (Fig. 2). In control wells, uTM preincubated with pre-immune IgG before reconstitution of Matrigel did not affect the uTM-induced inhibition of cell invasion. These results indicated that the inhibitory action of uTM on the cell invasion was not due to unknown materials in the preparation and was dependent on uTM. Furthermore, hirudin had no effect on the inhibitory action of uTM. The hirudin
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Fig. 1. Effect of uTM on invasion of B16F10 cells into Matrigel. The coated Matrigel (10 mg/®lter) was reconstituted with 100 ml of uTM (0, 10, 30, 100, 300, 1000 ng/ml) in serum-free medium containing 0.1% BSA for 1 h at room temperature. After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of uTM (®nal 0, 10, 30, 100, 300, 1000 ng/ml) were seeded onto ®lters in the upper surface of Cell Culture Insert. After 8 h incubation, the cells invading the lower surface were ®xed and assayed for invasion by using DiffQuick stain as described in Section 2. Experiments were repeated four times separately. Values are indicated as the mean ^ SD. *P , 0:05 compared with control, **P , 0:01 compared with control.
Fig. 2. Effect of hirudin or rabbit anti-human TM IgG on the inhibition of B16F10 cells invasion by uTM. Pre-incubation of uTM (®nal 100 ng/ml) with hirudin (®nal 0.1 unit/ml), rabbit anti-human TM IgG (®nal 100 mg/ml) or rabbit pre-immune IgG (®nal 100 mg/ml) was performed on ice for 30 min, and the solutions were added to the upper compartments 1 h before the cells were plated on the coated Matrigel (10 mg/®lter). After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of pre-treated uTMs (®nal 100 ng/ml) were seeded onto the ®lters. After 8 h incubation, the cells that invaded the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Experiments were repeated three times. Values are indicated as the mean ^ SD. **P , 0:01 compared with control.
235
was present at 0.1 unit/ml, a concentration suf®cient to completely inhibit the proteolytic/amidolytic activity of 0.01 NIH unit/ml thrombin. In control studies, there was no detectable thrombin amidolytic activity ,0:001 NIH unit/ml) nor detectable ecarin-activatable prothrombin in either the conditioned medium of cultured B16F10 cells (0:9 £ 105 cells), the membrane surface of cultured B16F10 cells (0:9 £ 105 cells) or in the Matrigel (0.1 mg/ml) (date not shown). Taken together, these results suggested that uTM repressed the invasion of B16F10 cells through Matrigel by thrombin-independent mechanism. To determine the structural requirements for this inhibitory action of uTM, the effect of various recombinant human soluble TMs on the invasion of B16F10 cells was compared with that of uTM (Fig. 3). Three recombinant soluble TMs, rTMa (67 kDa), rTMb (75±95 kDa), and rTME1±6 (40 kDa), were used in this study. Both rTMa and rTMb consist of the lectin-like, EGF-like and O-glycosylation site-rich domains of native TM, although rTMb has the chondroitin-4-sulfate glycosaminoglycan bound the O-glycosylation site-rich domain. rTME1±6 represents the EGF-like domain of native human TM. As shown in Fig. 3, Matrigel reconstituted with 1.3 nM rTMa or
Fig. 3. Effects of uTM and rTMs on invasion of B16F10 cells into Matrigel. The coated Matrigel (10 mg/®lter) was reconstituted with 100 ml of 1.3 nM soluble TMs (uTM, rTMa, rTMb and rTME1±6) in serum free medium containing 0.1% BSA for 1 h at room temperature. After removing the solution, B16F10 cells (2 £ 105 cells/100 ml) and 100 ml of each soluble TM (®nal 1.3 nM) were seeded onto the ®lters. After 8 h incubation, the cells on the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Each column represents the mean ^ SD of four values. *P , 0:05 compared with control.
236
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
rTMb signi®cantly decreased the invasion cells to 65± 70%, similar the Matrigel containing 1.3 nM uTM (100 ng/ml). In contrast, Matrigel containing 1.3 nM rTME1±6 did not inhibit the cell invasion. 3.2. Effect of uTM on the invasion of B16F10 cells in the presence of thrombin An earlier report demonstrated that high concentrations of thrombin accelerate adhesion of B16F10 cells to ®bronectin and metastasis of the cells [33]. Thus, the effects of uTM and/or hirudin on the invasion of B16F10 cells in Matrigel containing a high concentration of thrombin were investigated. Addition of thrombin (1 NIH unit/ml, approx. 10 nM) into Matrigel signi®cantly accelerated the invasion of B16F10 cells 1.5-fold when compared with the cell invasion level with the control Matrigel (Fig. 4). The thrombinenhanced invasion of B16F10 cells was decreased by addition of uTM (100 or 1000 ng/ml, approx. 1.3 or 13 nM) to the thrombin-containing Matrigel by 50% and 100%, respectively, relative to that observed with thrombin alone (Fig. 4). Addition of 10 units/ml
hirudin into the thrombin-contained Matrigel also completely abolished the accelerating action of thrombin on B16F10 cell invasion (Fig. 4). Moreover, co-addition of uTM (100 or 1000 ng/ml) with hirudin (10 units/ml) into the thrombin-containing Matrigel additionally enhanced the inhibitory activity (Fig. 4). These results indicated that while the thrombin-enhanced invasion activity was dependent on the proteolytic activity of thrombin, the inhibitory action of uTM was also exhibited in absence of thrombin proteolytic activity. 3.3. Effect of uTM on experimental lung metastasis of B16F10 cells The ability of uTM to inhibit experimental lung metastasis after intravenous injection of B16F10 cells in mice was investigated (Fig. 5). In this model, 11.6 ^ 1.9 colonies were observed per lung in control mice at 14 days after injection with B16F10 cells. In the case of single co-injection of uTM (1.74 mg uTM/kg of body weight) with B16F10 cells, the metastatic colonies of B16F10 cells in the lung was suppressed to 6.4 ^ 3.6 colonies/lung, but the reduction was not statistically signi®cant. After co-injection of 1.74 or 5.21 mg uTM/kg with B16F10 cells, additional injection of same amount of uTM, once a day up to 2 days (total three times), signi®cantly inhibited colony formation of B16F10 cells in the lung to 42 and 35%, respectively, of the colonies found in the control mice. 4. Discussion
Fig. 4. Effect of uTM on thrombin-stimulated B16F10 cell invasion. Pre-incubation of uTM (®nal 100, 1000 ng/ml) or hirudin (®nal 10 units/ml) with thrombin (®nal 1 NIH unit/ml) was performed for 30 min at 378C, and the solutions were added to the upper compartments 1 h before the cells were plated on the coated Matrigel (10 mg/®lter). After removing the solution, B16F10 cells (2 £ 105 cells/ 100 ml) and 100 ml of uTM (®nal 100, 1000 ng/ml) or hirudin (®nal 10 units/ml) pre-incubated with thrombin (®nal 1 NIH unit/ml) were seeded onto the ®lters. After 8 h, the cells invading the lower surface were ®xed and assayed using Diff-Quick stain as described in Section 2. Each column represents the mean ^ SD of four values. *P , 0:05 compared with control, **P , 0:01 compared with control, ²P , 0:01 compared with thrombin alone.
The present study indicated that Matrigel containing exogenous uTM inhibits invasion of B16F10 cells into the reconstituted basement membrane in vitro (Fig. 1). It is known that TM inhibits the procoagulant activity of thrombin and acts as a cofactor of thrombin-catalyzed protein C and TAFI activations and thrombin-catalyzed inactivation of single chain uPA. Therefore, we investigated whether the inhibitory action of uTM on tumor cell invasion was dependent on thrombin. No detectable activity of prothrombin/ thrombin was observed in B16F10 cells and Matrigel, and the inhibitory effect of uTM on the cell invasion was not interfered even in the presence of hirudin, a thrombin inhibitor (Fig. 2). Additionally, the throm-
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
Fig. 5. Effect of uTM on experimental lung metastases in mice. (A) B16F10 cells (1 £ 105 cells) were injected with or without uTM (0, 0.521, 1.74, or 5.21 mg/kg) into the lateral tail vein of BDF1 mice. Various dose of uTM were injected i.v. once a day for 2 days. A single administration of uTM (1.74 mg/kg) was coinjected with the cells i.v.. After 14 days, the mice (n 7) were sacri®ced and the lungs were removed. The lungs were ®xed and observed for metastasis as described in Section 2. Values are indicated as the mean ^ SD. *P , 0:05 compared with control, **P , 0:01 compared with control. (B,C) Representative photographs of mouse lung treated with or without uTM. (B) Control; (C) mice treated with 3 injections of uTM (5.21 mg/kg). Black
bin-enhanced invasion of B16F10 cells was also reduced by co-existence of uTM and hirudin in Matrigel to less than the invasion level observed in the control Matrigel (Fig. 4). These results suggest that the inhibitory action of uTM on cell invasion of B16F10 cells is due to a thrombin-independent mechanism. Tumor cell invasion consists of at least three steps,
237
including attachment of tumor cells to matrix, local degradation of matrix by tumor associated proteases and locomotion of tumor cells into tissues [1±6]. Rabhi-Sabile et al. [34] has reported that the chondroitin-4-sulfate of TM is an important moiety in cytoadhesion between Plasmodium falciparum-infected erythrocytes and TM-expressed cells, suggesting that the chondroitin-4-sulfate of TM plays an important role in interaction with cells. In addition, it has been noted that large chondroitin sulfate proteoglycans inhibited cell-matrix interaction [35,36]. Therefore, we examined the effect of chondroitin-4-sulfate of TM on B16F10 cell invasion by comparison of the inhibitory effects of rTMa and rTMb. The present data showed that the intensity of inhibitory effect of rTMb containing chondroitin-4-sulfate were similar to that of rTMa lacking chondroitin-4-sulfate (Fig. 3). Therefore, the inhibitory action of the soluble TMs on B16F10 cell invasion did not depend on the chondroitin-4-sulfate glycosaminoglycan attached to the O-glycosylation site-rich domain of the soluble TMs. Imada et al. [37] has noted that fetomodulin, which is identical to TM, appeared during differentiation of mouse teratocarcinoma F9 cells (an embryonal carcinoma stem cell line) into parietal endoderm cells, and suggested that fetomodulin/TM expressed on the surface of differentiated cells may act as a multifunctional protein during cell±cell interactions. It has further been postulated that the lectin-like domain of TM expressed on tumor cell surfaces could interact with cellular membranes and an extracellular matrix protein and may act as an adhesion molecule [22,24]. Interestingly, rTME1±6 lacking lectin-like domain did not inhibit B16F10 cell invasion (Fig. 3). The result implied that the lectin-like domain of soluble TM might contribute to interfere with attachment and/or locomotion of B16F10 cells to matrix through the interaction with some molecule in Matrigel or on the cell surface in the invasion process. However, we could not prove the mechanism in detail, because we can not yet examine the effect of lectin-like domain of soluble TM on B16F10 cell invasion. Additional studies are necessary to evaluate whether the lectin-like domain participates in the inhibitory action of soluble TM on tumor cell invasion. Previous papers reported that several inhibitors against Matrigel invasion of murine melanoma,
238
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
including B16F10 cells, also suppressed experimental lung metastasis in mice [36,38]. Therefore, it was possible that uTM may have similar effects in this murine model. Our experimental metastasis model in this study was mild compared with that in previous reports with respect to murine melanoma [36,38], because the metastatic colonies in control mice at 14 days after injection of B16F10 cells were about 12 colonies/lung. In this model, a single co-injection of 1.74 mg uTM/kg body weight with B16F10 cells showed a tendency to suppress the lung metastasis of the cells without statistical signi®cance, while additional injections of uTM, once a day for up to two days, more strongly and signi®cantly inhibited the lung metastasis (Fig. 5). It has been reported that micrometastasis is observed at several days postinjection of tumor cells in experimental lung metastasis model [1]. Therefore, the repeat injections of uTM for signi®cant inhibition of the lung metastasis may be required to maintain adequate plasma concentration of uTM in the invasion process of the tumor cells from plasma to extracellular matrix, although we do not yet know the half-life of uTM in the plasma of the mice injected with uTM. Most of tumor cells are capable of generating thrombin through activation of plasma coagulation cascade by tissue factor and/or tumor-associate proteases expressed on cell surface [39]. Esumi et al. [40] have demonstrated that injection of recombinant hirudin, a thrombin inhibitor, signi®cantly inhibited experimental lung metastasis of B16F10 cells, and concluded that the inhibition was related with inhibition of coagulation events by the inhibitor. In addition, the metastatic potential of B16F10 cells were modulated by uPA secreted from the cells [41,42]. Although the present in vitro experiments suggested that uTM inhibited Matrigel invasion of B16F10 cells in a thrombin-independent mechanism, we can not exclude a possibility that the inhibitory action of injected uTM on experimental lung metastasis of B16F10 cells in vivo may be resulted from inhibition of coagulation and/or ®brinolysis events in circulating plasma through the thrombin-dependent mechanism other than the thrombin-independent mechanism, since circulating plasma contains various proenzymes and active enzymes such as protein C, TAFI and uPA which are activated or inactivated by thrombin±TM complex.
The present study indicated that Matrigel containing uTM, puri®ed from human urine, inhibited the invasion of B16F10 cells in vitro by a thrombin-independent mechanism and intravenous injection of uTM suppressed tumor metastasis in an experimental metastasis model in mice. It remains the possibility that TM expressed on tumor cell surfaces also may regulate the invasion and metastasis of tumor cells. Experiments are now in progress to elucidate the detailed mechanism of suppression of invasion and metastasis of tumor cells by TM. Acknowledgements This work was partly supported by Grant-in-Aid for Encouragement of Young Scientists from The Ministry of Education, Science, Sports and Culture Japan Society for the Promotion of Science (grant nos. 09771997 and 11771450) and by the Sasagawa Scienti®c Research Grant from The Japan Science Society (grant no. 11-180). References [1] I.J. Fidler, Tumor heterogeneity and the biology of cancer invasion and metastasis, Cancer Res. 38 (1978) 2651±2660. [2] L.A. Liotta, Tumor invasion and metastasis: role of the extracellular matrix: Rhoads memorial award lecture, Cancer Res. 46 (1986) 1±7. [3] V.P. Terranova, E.S. Hujanen, G.R. Martin, Basement membrane and the invasive activity of metastatic tumor cells, J. Natl. Cancer Inst. 77 (1986) 311±316. [4] E. Bastida, A. Ordinas, Platelet contribution to the formation of metastatic foci: the role of cancer cell-induced platelet activation, Haemostasis 18 (1988) 29±36. [5] R.H. Kramer, G.L. Nicolson, Interaction of tumor cells with vascular endothelial cell monolayers: a model for metastatic invasion, Proc. Natl. Acad. Sci. USA 76 (1979) 5704±5708. [6] D.G. Menter, B.W. Steinert, B.F. Sloane, N. Gundlach, C.Y. O'Gara, L.J. Marnett, C. Diglio, D. Walz, J.D. Taylor, K.V. Honn, Role of platelet membrane in enhancement of tumor cell adhesion to endothelial cell extracellular matrix, Cancer Res. 47 (1987) 6751±6762. [7] C.T. Esmon, The roles of protein C and thrombomodulin in the regulation of blood coagulation, J. Biol. Chem. 264 (1989) 4743±4746. [8] K. Suzuki, H. Kusumoto, Y. Deyashiki, J. Nishioka, I. Maruyama, M. Zushi, S. Kawahara, G. Honda, S. Yamamoto, S. Horiguchi, Structure and expression of human thrombomodulin, a thrombin receptor on endothelium acting as a cofactor for protein C activation, EMBO J. 6 (1987) 1891±1897.
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240 [9] D. Wen, W.A. Dittman, R.D. Ye, L.L. Deaven, P.W. Majerus, J.E. Sadler, Human thrombomodulin: complete cDNA sequence and chromosome localization of the gene, Biochemistry 26 (1987) 4350±4357. [10] M. Zushi, K. Gomi, S. Yamamoto, I. Maruyama, T. Hayashi, K. Suzuki, The last three consecutive epidermal growth factor-like structures of human thrombomodulin comprise the minimum functional domain for protein C-activating cofactor activity and anticoagulant activity, J. Biol. Chem. 264 (1989) 10351±10353. [11] M. Tsiang, S.R. Lentz, J.E. Sadler, Functional domains of membrane-bound human thrombomodulin. EGF-like domains four to six and the serine/threonine-rich domain are required for cofactor activity, J. Biol. Chem. 267 (1992) 6164±6170. [12] H. Ishii, P.W. Majerus, Thrombomodulin is present in human plasma and urine, J. Clin. Invest. 76 (1985) 2178±2181. [13] H. Ishii, H. Uchiyama, M. Kazama, Soluble thrombomodulin antigen in conditioned medium is increased by damage of endothelial cells, Thromb. Haemost. 65 (1991) 618±623. [14] M. Nakano, M. Furutani, S. Hiraishi, H. Ishii, Characterization of soluble thrombomodulin fragments in human urine, Thromb. Haemost. 79 (1998) 331±337. [15] D.E. Jackson, T.J. Teras, H.H. Salem, C.A. Mitchel, Puri®cation and characterization of two form of soluble TM from human urine, Eur. J. Biochem. 221 (1994) 1079±1087. [16] Y. Takahashi, Y. Hosaka, H. Niina, K. Nagasawa, M. Naotsuka, K. Sakai, A. Uemura, Soluble thrombomodulin puri®ed from human urine exhibits a potent anticoagulant effect in vitro and in vivo, Thromb. Haemost. 73 (1995) 805±811. [17] Y. Takahashi, Y. Hosaka, K. Imada, T. Adachi, H. Niina, M. Watanabe, H. Mochizuki, Human urinary soluble thrombomodulin (MR-33) improves disseminated intravascular coagulation without affecting bleeding time in rats: comparison with low molecular weight heparin, Thromb. Haemost. 77 (1997) 789±795. [18] H. Wakabayashi, S. Natsuka, T. Mega, N. Otsuki, M. Isaji, M. Naotsuka, S. Koyama, T. Kanamori, K. Sakai, S. Hase, Novel proteoglycan linkage tetrasaccharides of human urinary soluble thrombomodulin, SO4-3GlcAb1±3Galb1-3(1/2Siaa2± 6)Galb1±4Xyl, J. Biol. Chem. 279 (1999) 5436±5442. [19] G.A.W. de Munk, E. Groeneveld, D.C. Rijken, Acceleration of the thrombin inactivation of single chain urokinase-type plasminogen activator (pro-urokinase) by thrombomodulin, J. Clin. Invest. 88 (1991) 1680±1684. [20] L. Bajzar, J. Morser, M.E. Nesheim, TAFI, or plasma procarboxypeptidase B, couples the coagulation and ®brinolytic cascades through the thrombin-thrombomodulin complex, J. Biol. Chem. 271 (1996) 16603±16608. [21] Y. Hosaka, Y. Takahashi, H. Ishii, Thrombomodulin in human plasma contributes to inhibit ®brinolysis through acceleration of thrombin-dependent activation of plasma procarboxypeptidase B, Thromb. Haemost. 79 (1998) 371±377. [22] T. Suehiro, M. Shimada, T. Matsumata, A. Taketomi, K. Yamamoto, K. Sugimachi, Thrombomodulin inhibits intrahepatic spread in human hepatocellular carcinoma, Hepatology 21 (1995) 1285±1290. [23] S. Wilhelm, M. Schmitt, J. Parkinson, W. Kuhn, H. Graeff,
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31] [32] [33]
[34]
[35]
239
O.G. Wilhelm, Thrombomodulin, a receptor for the serine protease thrombin, is decreased in primary tumors and metastases but increased in ascitic ¯uids of patients with advanced ovarian cancer FIGO IIIc, Int. J. Oncol. 13 (1998) 645± 651. M. Tabata, K. Sugihara, S. Tonezawa, S. Yamashita, I. Maruyama, An immunohistochemical study of thrombomodulin in oral squamous cell carcinoma and its association with invasive and metastatic potential, J. Oral. Pathol. Med. 26 (1997) 258±264. S.J. Kim, E. Shiba, H. Ishii, T. Inoue, T. Taguchi, Y. Tanji, Y. Kimoto, M. Izukura, S. Takai, Thrombomodulin is a new biological and prognostic marker for breast cancer: an immunohistochemical study, Anticancer Res. 17 (1997) 2319± 2323. H. Ishii, M. Nakano, J. Tsubouchi, T. Ishikawa, H. Uchiyama, S. Hiraishi, C. Tahara, Y. Miyajima, M. Kazama, Establishment of enzyme immunoassay of human thrombomodulin in plasma and urine using monoclonal antibodies, Thromb. Haemost. 63 (1990) 157±162. K. Nawa, K. Sakano, H. Fujiwara, Y. Sato, N. Sugiyama, T. Teruuchi, M. Iwamoto, Y. Marumoto, Presence and function of chondroitin-4-sulfate on recombinant human soluble thrombomodulin, Biochem. Biophys. Res. Commun. 171 (1990) 729±737. H. Hamada, H. Ishii, K. Sakyo, S. Horie, K. Nishiki, M. Kazama, The epidermal growth factor-like domain of recombinant human thrombomodulin exhibits mitogenic activity for swiss 3T3 cells, Blood 86 (1995) 225±233. A. Albini, H. Iwamoto, H.K. Kleinman, G.R. Martin, S.A. Aaronson, J.M. Kozolowski, R.N. McEwan, A rapid in vitro assay for quantitating the invasive potential of tumor cells, Cancer Res. 47 (1987) 3239±3245. L.M. Bafetti, T.N. Young, Y. Itoh, M.S. Stack, Intact vitronectin induces matrix metalloproteinase-2 and tissue inhibitor of metalloproteinases-2 expression and enhanced cellular invasion by melanoma cells, J. Biol. Chem. 273 (1998) 143±149. Q. Wu, M. Tsiang, S.R. Lentz, J.E. Sadler, Ligand speci®city of human thrombomodulin, J. Biol. Chem. 267 (1992) 7083± 7088. T. Morita, S. Iwanaga, Prothrombin activator from Echis carinatus venom, Methods Enzymol. 80 (1981) 303±311. M.L. Nierodzik, K. Chen, K. Takeshita, J.J. Li, Y.Q. Huang, X.S. Feng, M.R. D'Andrea, P. Andrade-Gordon, S. Karpatkin, Protease-activated receptor 1 (PAR-1) is required and ratelimiting for thrombin-enhanced experimental pulmonary metastasis, Blood 92 (1998) 3694±3700. S. Rabhi-Sabile, M. Steiner-Mosonyi, S. Pollefeyt, D. Collen, B. Pouvelle, J. Gysin, M.C. Boffa, E.M. Conway, Plasmodium falciparum-infected erythrocytes: a mutational analysis of cytoadherence via murine thrombomodulin, Thromb. Haemost. 81 (1999) 815±821. N. Sugiura, K. Sakurai, Y. Hori, S. Karasawa, S. Suzuki, K. Kimura, Preparation of lipid-derivatized glycosaminoglycans to probe a regulatory function of the carbohydrate moieties of proteoglycans in cell±matrix interaction, J. Biol. Chem. 268 (1993) 15779±15787.
240
Y. Hosaka et al. / Cancer Letters 161 (2000) 231±240
[36] K. Karasawa, N. Sugiura, Y. Hori, S. Suzuki, J. Onaya, K. Sakurai, K. Kimata, Inhibition of experimental metastasis and cell adhesion of murine melanoma cells by chondroitin sulfate-derivatized lipid, a neoproteoglycan with anti-cell adhesion activity, Clin. Exp. Metastasis 15 (1997) 83±93. [37] S. Imada, H. Yamaguchi, M. Nagumo, S. Katayanagi, H. Iwasaki, M. Imada, Identi®cation of fetomodulin, a surface marker protein of fetal development, as thrombomodulin by gene cloning and function assays, Dev. Biol. 140 (1990) 113± 122. [38] T. Mishima, J. Murata, M. Toyoshima, H. Fujii, M. Nakajima, T. Hayashi, T. Kato, I. Saiki, Inhibition of tumor invasion and metastasis by calcium spirulan (Ca-SP), a novel sulfated polysaccharide derived from a blue-green alga, Spirulina platensis, Clin. Exp. Metastasis 16 (1998) 541±550.
[39] L.R. Zacharski, V. Costantin, Blood coagulation activation in cancer: challenges for cancer treatment, HaÈemostaseologie 15 (1995) 14±20. [40] N. Esumi, D. Fan, I.J. Fidler, Inhibition of murine melanoma experimental metastasis by recombinant desulfatohirudin, a highly speci®c thrombin inhibitor, Cancer Res. 51 (1991) 4549±4556. [41] V.J. Hearing, L.W. Law, A. Corti, E. Appella, F. Blasi, Modulation of metastatic potential by cell surface urokinase of murine melanoma cells, Cancer Res. 48 (1988) 1270± 1278. [42] H. Yu, R.M. Schultz, Relationship between secreted urokinase plasminogen activator activity and metastatic potential in murine B16 cells transfected with human urokinase sense and antisense genes, Cancer Res. 50 (1990) 7623±7633.